Test to datetime null to NaT

[ryangilmour]

• closes BUG: pd.to_datetime() raises InvalidIndexError with Null-like arguments #35888
• tests added / passed
• Ensure all linting tests pass, see here for how to run them
• whatsnew entry

[mroeschke]

Looks like there's typing errors:

pandas/tests/tools/test_to_datetime.py:1074: error: List item 0 has incompatible type "None"; expected "NaTType"  [list-item]

[mroeschke]

Mind using less values here? (20/40 should be good)

[jreback]

don't we have a predefined breakpoitn with 50 values?

[mroeschke]

Yeah, but IIRC that's needed when specifically testing caching behavior

[ryangilmour]

@mroeschke - thanks for the review. I'll reduce the size of the test series. I've now got the dev environment set up to run mypy locally now. But I'm unsure how to change the code to avoid typing errors.

Mypy complains with either of the following Series objects because they have an incompatible type.

Series([NaT] * 20 + [None] * 20, dtype="object")
Series([None] * 20 + [NaT] * 20, dtype="object")

Possible fixes:

• use # type: ignore
  • not seen this in any other tests, so doesn't seem appropriate
• ignore the test case
  • is it not a valid test case?
  • it is a reasonable/possible input to the to_datetime() function, so I feel it ought to be tested
• something else?

[mroeschke]

I would just use # type: ignore[list-item] for now, and if you're interested you can look into adjusting the typing of to_datetime in a follow up PR.

[jreback]

I think we DO need to test a larger input (as well) here e.g. 'It only occurs if the input is large enough to force the caching mechanism.'

[jreback]

@ryangilmour can you merge master and address comments

[ryangilmour]

Yep - i'll get round to this in the next few days.

[jreback]

thanks @ryangilmour